Model-based analysis and evaluation of sensor faults in heat source system

Abstract

Faults cause building systems to under-perform in operation. An operation and maintenance process of building systems using fault detection and diagnosis (FDD) essentially requires evaluating the influence of faults in advance because deciding how to respond to faults is necessary to establish the strategies. However, many studies on FDD in a building’s operation have not considered these processes. Then, we focused on sensor errors as faults at ten temperature sensors and four flow sensors in a real heat source system and examined to evaluate the influence of faults. The real heat source system with two chillers and two cooling towers were used to demonstrate calculating system behavior with sensor faults, analyzing results, and evaluating the influence of sensor faults. We developed a detailed simulation model of the system covering a sensor network and combining automatic control system based on the specifications. Using this simulation, we calculated the system behavior without faults and behaviors with fourteen sensor faults in six fault severity levels. As for the annual system coefficient of performance (SCOP), the results showed that sensor faults had influence in various degree and each sensor has different features against fault severities. Some sensor faults had no effect on the system controls, but others had wide influence on controls of sub-systems and energy efficiency reduction. By considering these features of sensor faults, we evaluated the influence of each faults using the annual SCOP as an indicator. Using this method, we analyzed the influence of faults in detail and prioritized sensor faults using the indicator. It is expected that this fault evaluation method helps operation and maintenance of the system.